Isn’t llama.cpp already depending on Python anyways for the templating?
It uses a cpp implementation of jinja https://github.com/google/minja
It uses a cpp implementation of jinja https://github.com/google/minja